HRNet-Facial-Landmark-Detection
HRNet-Facial-Landmark-Detection copied to clipboard
Why do you re-write such "crop" function ?
First thanks for such great work. I found the crop function here is so different from the original code here. What is the purpose for such changes. As our own data would go here and failed to get properly cropped image finally.
Hi @MengHao666 : Were you able to figure out what the bug was? I think I have the same question as I face this issue #56. Everything seems to be working if I copy the old crop function you linked to (or crop_v2 in current branch). Is this the same for you?
Interestingly, the problem doesn't occur on all the images that go here though, and I'm not able to figure out what is wrong.